Machine learning photodynamics decode multiple singlet fission channels in pentacene crystal

Area of Science:

• Solid-state physics
• Photochemistry
• Materials science

Background:

• Crystalline pentacene exhibits high quantum efficiencies (>100%) due to ultrafast singlet fission.
• The precise mechanism of singlet fission in pentacene crystals remains debated due to experimental and computational limitations.

Purpose of the Study:

• To elucidate the competing singlet fission mechanisms in crystalline pentacene.
• To understand the interplay between electronic structure and molecular vibrations in excited-state dynamics.

Main Methods:

• A multiscale, multiconfigurational approach combined with machine learning photodynamics was employed.
• Simulations focused on realistic crystal dynamics, analyzing charge-transfer and coherent pathways in different dimer configurations.

Main Results:

• Simulations revealed coexisting charge-transfer-mediated and coherent singlet fission mechanisms in herringbone and parallel dimers.
• Predicted singlet fission time constants (61 and 33 fs) closely matched experimental values (78 and 35 fs).
• Intermolecular stretching was identified as critical for multi-exciton state generation and explaining anisotropy.

Conclusions:

• The study resolves the debate on singlet fission mechanisms in crystalline pentacene.
• Machine learning photodynamics enabled atomistic simulations of excited-state dynamics with high quantum mechanical accuracy.
• Findings provide a deeper understanding of light-harvesting processes in organic materials.